J. Phys. Chem. B 2010, 114, 6377–6385
6377
Model Studies of Membrane Disruption by Photogenerated Oxidative Assault Michael C. Howland† and Atul N. Parikh*,‡ Department of Chemical Engineering & Materials Science and of Applied Science, UniVersity of CaliforniasDaVis, DaVis, California 95616 ReceiVed: March 30, 2010
We have investigated the response of solid-supported phospholipid bilayers to short doses of photogenerated oxidative stress to characterize physical membrane changes during early phases of membrane oxidation. The low-dose oxidative stress is generated by uniformly exposing the bilayer samples using short-wavelength UV radiation (184-257 nm) for short periods (∼3 min) and resulting membrane morphological transformations characterized using a combination of wide-field epifluorescence microscopy and imaging ellipsometry measurements. Our results establish that the early phase of membrane oxidation is characterized by the nucleation and growth of discrete microscopic voids within the bilayer. The locations of the voids are randomly distributed throughout the sample surface, despite the uniform illumination. Over longer time scales, the voids continue to grow after the termination of the UV radiation. We also find that the voids heal as sample temperature is raised and that the supported bilayers consisting of fully saturated lipids are less susceptible to the mild oxidation conditions used, regardless of phase state. Analyzing these results in terms of (1) reactiveoxygen species mediated oxidative attack, (2) in situ generation of membrane oxidation products, and (3) their reequilibration between the membrane and the bulk aqueous phase explains the membrane morphological changes observed and provides insights into membrane perturbations following oxidative assault. Specifically, molecular properties of oxidation products (e.g., intrinsic curvature) account for formation and stabilization of voids within contiguous bilayers, and the long-term structural evolution is consistent with slow kinetics of the desorption of these oxidation products from the bilayer into bulk solution. A corollary benefit from our study is that the thermal properties of voids appear to offer a useful means to measure the thermal expansivity of supported membranes. Introduction Oxidation of membrane lipids is thought to play a role in a variety of pathological conditions such as apoptosis,1,2 aging,3,4 and Alzheimer’s disease5 via damage to the cellular membrane. Extensive experimental efforts using model membranes (e.g., vesicles) of defined chemical compositions and physical characteristics establish that the lipid oxidation-induced membrane damage is caused by significant alteration in the membrane structure and its physical properties. These studies further establish that structural changes including those in the molecular packing of membrane components,6,7 lateral fluidity,8 domain organization9-11 or molecular redistributions,12 membrane permeability,13 and phase behavior following lipid oxidation all play a role. The exact relationships between the chemical changes induced by oxidation and the resultant structural perturbations of the membrane remain poorly understood. It is now well-established that the membrane oxidation generates a variety of partially oxidized lipids including peroxides, aldehydes, and carboxylic acids. The oxidation process changes the chemical composition of the lipids and affects many properties of the individual species. Many polyunsaturated lipids have been shown to have important roles in chemical signaling mechanisms in response to oxidative assault.14,15 In the current study, we focus on monounsaturated * Corresponding author. Applied Science Department, 3007 Engineering III, University of CaliforniasDavis, Davis, CA 95616. Phone: +1 (530) 304-7523. Fax: +1 (530) 752-2444. E-mail:
[email protected]. † Department of Chemical Engineering & Materials Science. ‡ Department of Applied Science.
lipids and investigate the structural response of the membrane. Monounsaturated species are less susceptible to oxidation; however, the physical properties of these species generated by oxidation are still expected to differ considerably from their lipid analogues. These differences are likely to play a major role in the membrane perturbation. Their chemical and structural differences raise many fundamental questions about the influence of these species on the properties of the membrane itself. First, the loss of a partial or complete tail results in a decreased area per molecule, likely changing the packing density of the membrane, which may affect a variety of properties such as lateral mobility. Second, the absence of this tail changes the shape of the molecule, changing the intrinsic curvature and potentially stabilizing nonplanar membrane structures. Third, the difference in reactivity between saturated and unsaturated lipids should result in different oxidative susceptibilities and thus different membrane responses to the initial reactive oxygen species (ROS) attack. Fourth, the reduction in the overall hydrophobic character increases the water solubility of these lysolipids compared to intact lipids, which may change the dynamics of molecular transport from the membrane into the bulk solution. Fifth, the differences in the physical character of the lysolipids may change the mobility and thermal properties of the membrane. Addressing how these changes affect membrane structure will afford a more complete picture of oxidation-induced membrane reorganizations. However, real-time studies of lysolipid exchange and membrane reorganization using the popular vesiclebased model systems have proved challenging because of experimental difficulties. To this end, planar supported lipid
10.1021/jp102861v 2010 American Chemical Society Published on Web 04/23/2010
6378
J. Phys. Chem. B, Vol. 114, No. 19, 2010
bilayer membranes offer an attractive model membrane configuration for the induction of membrane oxidation in situ. Supported lipid bilayers are typically formed at the solid-liquid interface when vesicular microphases of lipids and their mixtures rupture and spread spontaneously on hydrophilic surfaces.16,17 When appropriately formed, they are essentially separated from the substrate surface through an intervening cushion layer18,19 (e.g., hydration layer of water 5-15 Å thickness on silica surfaces) and exhibit two-dimensional contiguity and lateral fluidity reminiscent of lipid membranes of vesicles. Because of their planar substrate-confined character, they are amenable to detailed, real-time characterization using many typical surface science based techniques including fluorescence microscopy, imaging ellipsometry, and scanning probe microscopy.20 Recently, we have shown that photogeneration of oxidative stress provides a convenient means to introduce controlled proportions of oxidative stress to supported lipid bilayers.21 In particular, others and we have shown that an extended illumination by short-wavelength UV radiation (184-257 nm) in conjunction with a photomask results in local oxidation and removal of fluid supported bilayers from the exposed areas of the bilayer. An illumination time of >20 min by a mediumpressure Hg lamp (10-15 mW cm-2 intensity near the lamp surface) is required to completely remove the bilayer and afford micropatterning. For some unsaturated lipids, a direct photolysis due to direct resonance with p-p* transitions is possible. However, a significant body of literature suggests that the dominant effect is due to indirect oxidation by the photogenerated reactive oxidative species (ROS) in the aqueous phase.21 The latter include a combination of strong oxidizing agents, including hydroxyl radicals (oxidation potential, 2.80 V), singlet atomic oxygen (2.42 V), ozone (2.07 V), and hydrogen peroxide (1.88 V). Interactions of ROS on lipids have been extensively studied. These studies establish the formation of a complex cocktail of oxidants including short-chain peroxides, aldehydes, and carboxylic acids attached to the phospholipid headgroup as well as other small molecule byproducts that are part of what was formerly the phospholipid tail. In the work reported here, we investigate the macro- and microscale response of phospholipid bilayers after exposure to short doses (∼3 min) of photogenerated oxidative stress. These short durations of UV illumination create lower concentrations of ROS and offer an insight into the early phases of the membrane degradation pathway. We focus our investigation on the effects of the oxidation process on the residual membrane. We characterize the change in the membrane structure after exposure using a combination of epifluorescence microscopy and imaging ellipsometry techniques. We find that the early phase of membrane oxidation is characterized by the nucleation and growth of microscopic voids. The locations of the voids are spatially uncorrelated and appear randomly distributed, despite the initial uniform illumination. Interestingly, we also find that the voids continue to grow even after the termination of the UV radiation, suggesting long-term structural evolution consistent with slow kinetics of the loss of oxidation products from the bilayer. A comparative study of the responses from saturated lipids in fluid and gel phases with those from fluidphase unsaturated lipids reveals that saturated lipids are resistant to these mild oxidation conditions, regardless of phase state. We also investigate the lateral fluidity and thermal properties of the perforated membrane following oxidation. The observed changes in the membrane structure are considered in the context of changes in the packing density, intrinsic curvature, and solubility of the lipid species upon oxidation.
Howland and Parikh Materials and Methods Materials. 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC), 1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC), and naturally derived chicken egg L-R-phosphatidylcholine (EggPC) were obtained from Avanti Polar Lipids (Alabaster, AL). Texas Red 1,2-dihexadecanoyl-sn-glycero-3-phosphoethanolamine, triethylammonium salt (TR-DHPE), was obtained from Invitrogen (Carlsbad, CA). All lipids were suspended and stored in chloroform in the freezer (-20 °C) until use. Hydrogen peroxide (30% V/V) and sulfuric acid were purchased from J. T. Baker (Phillipsburg, NJ) and Fisher Chemicals (Fairlawn, NJ), respectively, and used as received. All chemicals were used without further purification. Organic-free deionized water of high resistivity (∼18.2 MWcm) was obtained by processing water first through a reverse osmosis deionization unit and then a Millipore Milli-Q filtration unit (Billerica, MA). Phosphate buffer saline (PBS, PH ) 7.2, 154 mM NaCl, 1.54 mM KH2PO4, and 2.71 mM Na2HPO4) was obtained from Invitrogen and used as a vesicle spreading solution. Glass coverslips were purchased from Corning (Corning, NY). Silicon substrates with native oxide overlayers were purchased from Silicon Sense (Nashua, NH). Silicon substrates with nominally 1000 Å of thermally grown oxide were purchased from Silicon Quest International (Santa Clara, CA). Substrate Preparation. Substrates (silicon wafers or Corning glass coverslips) were cleaned from adventitious contaminants by oxidizing in a freshly prepared 4:1 (v/v) mixture of sulfuric acid and hydrogen peroxide for a period of 5 min maintained at ∼100 °C (Caution: this mixture reacts violently with organic materials and must be handled with extreme care). The substrates were then withdrawn using Nalgene tweezers and rinsed immediately with a copious amount of deionized water. Substrates were stored under water and dried in a stream of nitrogen just before use. All cleaned, oxidized substrates were used within 1 day of the pretreatment. Small Unilamellar Vesicle Formation. Small unilamellar vesicles (SUVs) were prepared using vesicle extrusion methods.22 A desired amount of lipid or lipid mixtures suspended in chloroform were mixed in a glass vial. The solvent phase was then evaporated under a stream of nitrogen and subsequently evacuated for at least 2 h in a vacuum desiccator. The dried lipid mixture was then suspended in Millipore water to be rehydrated. The total lipid concentration was 2 mg/mL. Dye concentrations were 1 mol % for stocks containing TR-DHPE. The desired amount of hydrated aqueous solution was then sonicated and passed through a Avanti Mini-Extruder (Avanti, Alabaster, AL) using 0.1 um polycarbonate membrane filters (Avanti, Alabaster, AL) 21 times at a desired temperature (at least 10 °C above the transition temperature for the major lipids). One part of the resulting SUV solutions was diluted with one part of PBS before use. The SUV solutions were used within a few hours of extrusion. Supported Bilayer Formation. Supported phospholipid bilayers were formed using the previously reported vesicle fusion and rupture method.18,23,24 Vesicle spreading was carried out by placing the substrates over a ∼50 µL SUV drop placed at the bottom of a crystallization well. The samples were allowed to incubate for ∼20 min to ensure equilibrium coverage. The well was then filled with water and transferred to a large reservoir of water in which the substrate was shaken gently to remove excess lipids. Finally, the substrates were transferred to culture wells making sure that the substrates remain sub-
Membrane Disruption by Oxidative Assault
Figure 1. Schematic diagram of the UV exposure setup used in these experiments. Samples are exposed to short-wavelength UV radiation generated by a quartz-enveloped mercury grid lamp. The light travels ∼26 mm (including ∼5 mm through water) before reaching the membrane. The sample is held in a small glass dish and mounted on a stack of microscope slides. Care is taken to avoid any direct air exposure of the sample throughout the experiment.
merged in water. Supported bilayer samples prepared in this way were then stored in deionized water for further characterization. Short-Wavelength UV Exposure. Substrates with supported bilayers were exposed to UV in a home-built exposure apparatus shown schematically in Figure 1. Short-wavelength UV radiation (187-254 nm) is generated using a Hg grid lamp at the top of the setup. The light travels ∼26 mm (including ∼5 mm through the overlying water layer) before contacting the membrane. The samples were held in a small glass dish and mounted on a stack of microscope slides to decrease the travel distance of the light. Care was taken to avoid any direct air exposure of the membrane throughout the process. Fluorescence Microscopy. A Nikon eclipse TE2000-S inverted fluorescence microscope (Technical Instruments, Burlingame, CA) equipped with an ORCA-ER (model LB10-232, Hamamatsu Corporation, Bridgewater, NJ) or Retiga-1300 CCD camera (Technical Instruments, Burlingame, CA) and a Hg lamp as the light source was used to visualize all fluorescent samples. Two filter wheels, one containing a set of excitation and the other emission filters, were mounted in front of the light source and the CCD camera, respectively. An extra triple-band emitter was installed in the dichroic mirror cube for aiding in focusing through the eyepiece. Typically, images taken were using a Plan 10X (NA, 0.25) and a 60X (NA, 0.7) objective. Images were stored and processed using SimplePCI software (Compix, Inc., Cranberry Township, PA). Excitation and emission maxima for the probes used were 583/601 nm for TR-DHPE. Imaging Ellipsometry. Ellipsometric angle measurements and spatially resolved ellipsometric contrast images were acquired using an Elli2000 imaging system (Nanofilm Technologie, Gottingen, Germany). The ellipsometer employs a frequency-doubled Nd:YAG laser (adjustable power up to 20 mW) at 532 nm and equipped with a motorized goniometer for an accurate selection of the incidence angle and corresponding detector positions. The ellipsometer employed the typical PCSA (polarizer-compensator-sample-analyzer) nulling configuration in which a linear polarizer (P) and a quarter-wave plate (C) yield an elliptically polarized incident beam. Upon reflection from the sample (S), the beam is gathered via an analyzer (A) and imaged onto a CCD camera through a long working distance 10X objective. The P, C, and A positions that yielded the null condition are then converted to the ellipsometric angles, Ψ and ∆. Measurements were generally taken at an incidence angle of 60°. Silicon substrates with a native oxide overlayer (SiO2/ Si) whose surface chemistry is comparable to that of glass were used to enhance the optical contrast with the lipid phase. For characterization under aqueous conditions, a fluid cell was used
J. Phys. Chem. B, Vol. 114, No. 19, 2010 6379 (Nanofilm Technologie, Go¨ttingen, Germany). The cell consisted of a Teflon chamber (2.6 mL volume) with glass windows fixed at 60° (incidence angle) to the substrate normal. To determine membrane thicknesses, an optical model is employed. The specific parameters of our model are detailed elsewhere.25 Briefly, the membrane is modeled as a single slab with a refractive index of 1.45 sitting atop a silicon substrate with an appropriately 2 nm thick silicon oxide layer. The above water layer is taken to be semi-infinite with a refractive index of 1.33. Image Analysis. Analysis of void size was conducted using ImageJ (NIH). Sixteen bit single-channel images in TIF format were imported from Simple PCI. A background subtraction was conducted using a rolling ball radius of 15 pixels to correct for the illumination profile. The intensity histogram was then normalized, and an intensity threshold was set at 49 000 (out of 65 536 for a 16 bit image) for particle selection. Voids were identified using the Analyze Particles function in ImageJ. Particles comprised of less than nine pixels were ignored to account for noise resulting from the threshold process. No constraints on circularity were imposed for particle selection. Results Membrane Reorganization Induced by UV-Illumination. Illumination of single supported phospholipid bilayers, comprised of 1% TR-DHPE and 99% POPC, using ozone-generating short-wavelength UV radiation (187-254 nm) is first examined using a combination of imaging techniques. Previous studies have established that epifluorescence microscopy data of such monocomponent lipid bilayers containing small amounts of probe produce uniform fluorescence intensity. To confirm the initial integrity of the membrane prior to UV exposure, each sample was first observed immediately after formation, and the fluorescence recovery after photobleaching (FRAP) was monitored.26 Samples were observed to exhibit uniform fluorescence emission and reasonable lateral fluidity as expected for healthy, intact membranes (data not shown). Figure 2A shows a fluorescence micrograph of a typical sample 90 min after a 3 min exposure to uniform UV radiation. Two noticeable differences are observed. First, we note that the fluorescence intensity of the sample is greatly reduced. This uniform diminution in fluorescence emission can be attributed to sample photobleaching presumably during UV illumination. For further data acquisition and analysis, we compensate for this intensity loss by increasing the exposure time used to collect the fluorescence data. Second, micrometer-sized regions devoid of fluorescence emission appear randomly scattered across the sample. Comparing the residual intensity of these features with that of a physical scratch confirms that the features exhibit little or no fluorescence emission (see Supporting Information). These fluorescence-free areas are as large as several micrometers (∼3 µm) in diameter. It is notable that while the UV illumination occurs uniformly over the entire sample surface the fluorescence-free regions appear as discrete spots that are randomly distributed and spatially uncorrelated. To further characterize these fluorescence-free features, we use imaging ellipsometry (IE). IE measures changes in the optical polarization state of the impinging laser beam upon reflection from the sample surface. Using classical electromagnetic theory in conjunction with a parallel-slab model for the ambient phase/bilayer film/substrate configuration, IE furnishes quantitative estimates for the local surface density or thickness of thin films (e.g., lipid bilayers).25 To facilitate quantitative IE characterization, samples were prepared on silicon wafers with ∼2 nm of native oxide following a sample preparation procedure
6380
J. Phys. Chem. B, Vol. 114, No. 19, 2010
Howland and Parikh
Figure 2. Supported POPC membrane containing 1% Texas Red DHPE 90 min after a 3 min exposure to short-wavelength UV radiation. Panel (A) shows a fluorescence micrograph of a sample on a transparent glass coverslip. The scale bar is 50 µm wide. Panel (B) shows a similarly formed sample on silicon, imaged by imaging ellipsometry. The scale bar is 100 µm wide. The optical configuration is such that a ellipsometric nulling condition is achieved for the bilayer (see text for details). Brighter areas indicate a lower density of lipid material. Occasional doubling and/or broadening of features is visible in the ellipsometric image due to scanning artifacts.
Figure 3. POPC membrane containing 1% Texas Red DHPE supported on a silicon possessing a thermally grown thick oxide layer. The sample is shown after exposure to 3 min of short-wavelength UV radiation. Panel (A) shows a fluorescence micrograph of a sample. Panel (B) shows the same sample imaged by imaging ellipsometry. The scale bars are 10 µm wide. The optics are set such that the nulling condition is achieved for the bilayer and brighter areas indicate a lower density of lipid material.
comparable to that used for samples prepared on glass substrates for the fluorescence measurements reported above. Figure 2B shows an ellipsometric micrograph of a typical sample. The optical configuration of IE has been adjusted such that dark areas result from ellipsometric parameters corresponding to a full bilayer. The brighter spots, in this configuration, represent areas of reduced bilayer density or thickness. Individual measurements taken at several locations across this image confirm the integrity of the residual bilayer and offer some insight into the properties of the bright features. Determining the average ellipsometric parameters of the bright features and their surroundings, and computing the corresponding thicknesses, suggests that the surroundings stand 5.2 ( 1.1 nm taller than the bright features. This optical thickness value agrees well with those reported for an intact POPC at room temperature.27 Moreover, the thickness contrast between the surrounding lipid and the bright features suggests that the latter are lipid-free voids. Due to limited lateral resolution wherein single void sizes are on the scale of only a few pixels, we cannot rule out the possibility that small amounts of material remain in the void regions. To establish the correspondence between the fluorescencefree features observed in FM and the physical voids seen in IE, a companion set of samples was prepared on silicon wafers with 100 nm of thermally grown oxide. Although the presence of the thick oxide reduces the precision with which ellipsometric thicknesses can be determined in our experimental geometry, the use of these substrates allows for simultaneous characterization of single bilayer samples using both FM and IE techniques. Figure 3 shows a sample, prepared as above, on a thick oxide silicon wafer, characterized by both FM (A) and IE (B) at random locations on the sample. The two images reveal comparable features. We can qualitatively confirm that the
discrete features observed in IE are of lower density than the surrounding area. These features correlate well in size and distribution to those observed in fluorescence in panel (A), confirming that the features observed using FM and IE represent the same physical perturbations to the otherwise uniform POPC bilayers. Additional evidence for the presence of lipid-free microscopic voids in UV-illuminated samples is provided by exposing the sample to new lipid material. The substrate exposed by the voids is accessible to adsorption by membrane-compatible species in solution. Exchanging the surrounding water for PBS buffer and introducing new lipid material as SUVs result in fusion of additional vesicles to the surface and backfilling of the voids (see Supporting Information) Many brighter regions are visible after exposure to new vesicles, indicating the successful addition of lipid material containing unbleached probes.28 Taken together, these results and those observed on glass and silicon (reported above) indicate that the spotlike features generated following UV illumination are membrane voids within the otherwise contiguous single lipid bilayer. To further characterize the physical properties of a membrane after oxidation, we examine the lipid and void mobility using FRAP. Figure 4 shows three frames selected from a sequence of fluorescence micrographs obtained for a typical sample after 3 min of UV exposure. They correspond to images acquired before, immediately following, and 12 min after the illumination by an intense photobleaching beam (see experimental). A comparison of these images reveals two key features. First, we see near-complete fluorescence recovery in the bleached areas of the sample encapsulating many voids indicating that the residual bilayer continues to exhibit long-range lateral fluidity even after the short UV illumination. Second, the relative
Membrane Disruption by Oxidative Assault
J. Phys. Chem. B, Vol. 114, No. 19, 2010 6381
Figure 4. Supported POPC membrane containing 1% Texas Red DHPE after exposure to 3 min of short-wavelength UV radiation. All panels show the same sample location, and the scale bars are 10 µm wide. Panel (A) shows the sample before photobleaching. Panel (B) shows the sample immediately after bleaching, and panel (C) shows it 12 min later confirming fluorescence photobleaching recovery.
Figure 5. Supported POPC membrane containing 1% Texas Red DHPE after exposure to 3 min of short wavelength undergoing temperature cycling. The scale bars are 10 µm wide. Sample (A) at room temperature; (B) at 36.7 °C after heating at >2 °C/min; (C) at room temperature after cooling at >2 °C/min; (D) at 37.6 °C after heating >2 °C/min; (E) at room temperature after cooling at 2 °C/min; and (G) at room temperature after cooling at >2 °C/min.
invariance in the locations of the voids suggests that they are essentially immobile over the time course of the experiment (∼12 min). Though a membrane containing such large immobile regions can be difficult to quantitatively analyze using typical microscopy-based FRAP analysis,29 we can use this approach to qualitatively estimate a lower bound for the lipid diffusivity after void formation. Using an in-house program written to measure the recoveries30 on a typical sample, we estimate the diffusivities to be 3.2 ( 0.7 and 3.8 ( 1.1 µm/s before and after void formation, respectively (see Supporting Information). The diffusivity of the remaining membrane is comparable to that of the unexposed membrane, within the margins of error the respective fits. This provides further verification of the unharmed nature of the remaining membrane. It is worth noting that in multiple repeated experiments qualitatively similar regions devoid of fluorescence intensity and density (as determined by IE) were observed; however, the size and distribution of these regions vary from day to day. These variations may straightforwardly be attributed to a number of uncontrolled factors. First, other uses of the UV lamp between sample exposures results in a change in source intensity over time. Second, thermal fluctuations during the UV exposure are likely to change the kinetics of the bilayer degradation. However, within a set of experiments conducted in immediate succession these factors can be reasonably controlled. Under such conditions, we find that the extent of oxidative damage and subsequent membrane restructuring is proportional to the UV exposure time. For three different samples exposed for three different extents, typical voids are observed in each case (see Supporting Information). However, the amount of sample surface occupied
by voids varies with the exposure time. Exposures of 60, 180, and 360 s result in 3.6, 12.9, and 26.9% coverage of voids. The amount of material loss is proportional to the UV illumination time, within the relatively large error of these experiments. Longer exposures substantially bleach the included dyes and make characterization of the samples impractical. However, previous studies have shown that 20 min of exposure in a similar experimental geometry results in a complete degradation of a POPC bilayer.31 With proper calibration, such exposure control can be used to tailor the size and coverage of the voids between minimal void occurrence and full bilayer removal. Thermal Properties of the Membrane after Restructuring. To examine the thermotropic behavior of voids, we thermally cycle UV-illuminated bilayers between 5 and 37 °C. Figure 5A shows a sample exhibiting voids obtained by UV-exposureinduced mild photo-oxidation. The sample is heated to 36.7 °C (Figure 5B). At this temperature, presumably due to thermally induced area expansion of the lipids, the voids heal and are no longer visible (within the resolution of the 60× objective). Upon subsequent cooling to room temperature at a rapid rate (>2 °C/ min), the voids reappear (Figure 5C). This thermally induced growth (during cooling) and shrinking (during heating) of voids occurs gradually (data not shown). This process can be repeated multiple times to cycle the void formation and healing. Many of the voids are observed to reform at locations at or near their original positions after cooling (see Supporting Information). The rate of temperature cycling can also result in changes in the size distribution of the voids. Figure 5D,E shows the results of heating a sample such that the voids heal over. The sample
6382
J. Phys. Chem. B, Vol. 114, No. 19, 2010
Howland and Parikh
Figure 6. Supported POPC membrane containing 1% Texas Red DHPE after exposure to 3 min of short-wavelength UV light monitored over time. Selected frames at (A) 0 s, (B) 813 s, and (C) 3100 s after transfer to the microscope are shown. The scale bar is 10 µm wide. Images (B) and (C) have had their brightness adjusted. The percent coverage as a function of time is shown in (D) along with a decaying exponential fit for a rate constant of 0.0012 s-1. The data corresponding to panels (A), (B), and (C) have been starred for clarity.
is held at an elevated temperature for >60 min and then cooled slowly (2 °C/ min) described above, voids are observed to reform when the temperature is lowered back to room temperature. However, the voids formed as a result of this slower cooling rate are significantly larger, having an average size of 3.4 µm2 per void compared to the average size of 1.1 µm2 per void from the faster cooling rate shown in Figure 5C. The increase in size per void is compensated for by a reduction in the number of voids across the sample surface, maintaining approximately the same area coverage before and after temperature cycling. This process is reversible as shown in Figure 5F,G. After heating the sample back to a temperature sufficient to heal over the voids, a fast cooling step (>2 °C/min) results in the reformation of small voids. The sizes are comparable to those seen after the first fast cooling cycle. Growth of Voids after Formation. Careful observation of the voids after formation reveals that their morphology is not completely set upon the termination of UV exposure. Figure 6 shows the evolution of a typical sample immediately after UV exposure. Panels (A), (B), and (C) show selected frames taken 0, 813, and 3100 s after transfer of the sample from the UV exposure apparatus to the fluorescence microscope stage, a process which takes ∼1-3 min. Immediately following UV exposure, the voids are clearly visible and occupy a significant portion of the sample surface. The voids continue to grow slowly for ∼90 min after the termination of UV illumination. Voids present in (A) increase in size, and additional voids appear by panel (B), though it is not possible to determine if these are entirely new voids or have just grown to a resolvable size. Panel (C) shows the void region morphology approaching an asymptote in total area and individual size. Panel (D) summarizes this effect by showing the percent coverage of lipid materials plotted against time. The lipid coverage area decreases steadily over time as the voids grow, eventually leveling off after more than 3000 s. The data points corresponding to panels (A)-(C) have been bolded in panel (D) for clarity. A decaying exponential
line of best fit (see Discussion below) is shown on the same plot. From the lower bound of this curve, we can determine the total amount of lipid loss for a given sample. The average loss over all of our experiments resulting from 3 min of UV exposure is 24 ( 11%. Effect of Lipid Phase and Saturation. To investigate the role of molecular properties of parent lipids in the UV-induced void formation and membrane reorganization, we compare the 3 min room temperature UV illumination of (1) POPC, a fluid phase lipid at room temperature (Tm ) -2 °C), with a monounsaturated acyl tail; (2) DMPC, a room-temperature gel phase lipid (Tm ) 24.0 °C), with fully saturated acyl tails; and (3) EggPC, a naturally derived lipid mixture containing approximately 45% saturated and 55% unsaturated lipids, in its fluid phase (Tm ∼0 °C) at room temperature. EggPC experiments were conducted at room temperature, well above the Tm of the mixture. Experiments with DMPC are performed at ∼10 and ∼35 °C straddling the Tm. For EggPC, voids appeared immediately after 3 min UV exposure and continued to grow over time. The morphology of the voids is similar to that of the features observed in POPC; however, in EggPC, the voids occupy a reduced area fraction of the surface, relative to POPC. For DMPC, no such voids were observed at either temperature after UV exposures of 3 min. Previous studies have shown that DMPC can be degraded by significantly longer (>20 min) exposures to short-wavelength UV.31 The absence of void formation above and below the Tm of DMPC in the current study suggests that the fluid phase state of the membrane alone is not sufficient to produce void formation after 3 min exposures. A comparison of the responses of DMPC with EggPC and POPC suggests that the presence of unsaturated acyl tails within the lipids comprising the membrane is necessary for void formation after such short exposures. Given the near or below 0 °C Tm of the two unsaturated lipids used here, we cannot comment on the response of unsaturated lipids in the gel phase. Taken together, these data suggest that the observed formation of voids
Membrane Disruption by Oxidative Assault may be a general response of fluid-phase unsaturated supported lipid membranes to short exposures of short-wavelength UV radiation. Discussion Results presented in this study suggests that controlled exposure to short-wavelength UV radiation provides a convenient model to probe effects of oxidative stress on lipid membranes in real-time. Exposure to membrane bilayers by UV radiation influences membrane bilayers via two distinct pathways: (1) direct breakdown of lipids by UV radiation and (2) oxidation of lipids by ROS created by UV radiation. We have recently shown that membrane damage resulting from shortwavelength UV radiation such as that used in the present study is significantly reduced in the presence of radical scavengers,32 supporting the notion that the primary mechanism of membrane disruption is that of lipid oxidation by ROS and not the action of direct UV photolysis. The ROS generated by the wavelengths used in this study include a cocktail of highly potent oxidants including ozone, singlet molecular oxygen, peroxides, and hydroxyl radicals.21 These ROS are known to attack the double bonds in unsaturated lipids and produce aldehydes and hydroxyhydroperoxides,33 which subsequently oxidize further, resulting in the formation of carboxylic acids.34 Our observations that bilayers comprised of unsaturated lipids (e.g., POPC and EggPC) readily reveal membrane disruption following exposure to low-dose short-wavelength UV radiation, and then fully saturated ones (e.g., DMPC) lend additional support to the ROSmediated pathway. Furthermore, these photogenerated ROS closely resemble the mixture of oxidants produced during many biological oxidative stress scenarios, suggesting a common mechanism of their production, presumably via water oxidation pathway.35 This close parallel further corroborates the suitability of photogenerated oxidative stress as a means to explore membrane oxidation. A central finding of our study is that low-dose exposure of single supported lipid bilayers to short-wavelength UV radiation perturbs the membrane structure by producing localized microscopic voids. Because short-wavelength UV radiation is known to produce a potent mixture of oxidants, short exposures used in the present study shed light on membrane response to lowdose oxidative stimulus which mimics early events during membrane oxidation. This formation of physical voids within the membrane indicates a loss of membrane molecules and an attendant restructuring of the membrane. Furthermore, our results reveal a continued growth of voids following the membrane photo-oxidation for a significant period of time indicating a long-term character of the structural reorganization. These observations shed light on the mechanism of membrane photo-oxidation such as discussed below. Mechanistic Considerations. The primary site for ROS attack of membrane lipids is double bonds producing fatty acid fragments which readily dissolve in the aqueous environment. However, during low-dose exposure and early oxidation events, a significant proportion of partially oxidized membrane-soluble components are also formed.36 These include short-chain, but amphiphilic, carboxylic acids, aldehydes, and lysolipids such as those formed by the loss of an entire acyl tail. Many of these membrane-soluble oxidation products can be expected to exhibit considerable differences in their structural properties from their parent phosholipids in a variety of ways. First, the area per molecule is considerably lower, resulting in changes to the lipid packing density.7 Second, the relative volumes of hydrophilic headgroup and the hydrophobic tail are substantially altered after
J. Phys. Chem. B, Vol. 114, No. 19, 2010 6383 partial or complete chain cleavage, thus exhibiting widely different intrinsic curvature in comparison to their parent unoxidized lipids.37 Third, the hydrophobic character of lipid tails is substantially reduced upon the addition of polar groups during oxidation. This addition in turn reduces the molecular affinity of the oxidation products for the membrane environment and increases their solubility in water. The structural reorganization including details of void formation and growth we observe following photo-oxidation of single supported lipid bilayers in the present study can be understood within the context of these changes and is discussed in detail below. First, the appearance of microscopic voids within the supported phospholipid bilayers immediately following the short exposure to deep UV radiation indicates a large-scale reduction in the membrane area. This areal shrinkage of the supported membrane can result from either (1) the loss of a fraction of lipids due to complete oxidation or (2) the generation of partially oxidized lipids. The former generates a short-chain fatty acid and other small molecular fragments, which readily dissolve in the aqueous phase reducing net membrane density. The latter partially oxidized species exhibit an increased solubility in water but are also expected to partition into the membrane environment. Because of the reduced area per molecule of these species, their formation can also contribute to the void formation. From our present experiments alone, it is not possible to fully quantify the relative contributions of the two processes. To separate these two effects, probes that differentiate between the loss of lipid tails and loss of head groups (e.g., vibrational spectroscopy markers) are needed. Such experiments are currently in progress. It is notable that the voids appear randomly distributed across the supported bilayer surface despite uniform illumination of the entire bilayer surface. It is now known that fluid phospholipid bilayers such as those used in the present study exhibit a wide range of lateral density fluctuations. Indeed, significant evidence now establishes that in-plane translational mobilities in lipid bilayers couple with out-of-plane bilayer undulations to produce a range of overlapping collective membrane motional modes.38 These dynamical modes transiently generate regions of differing molecular densities. While the membrane interactions with the strongly adhesive silica substrate are believed to suppress some of these collective membrane modes, it is known that wellformed bilayers retain at least some of these dynamical features of free membranes.39,40 The regions of reduced membrane density must increase solvent (and thus ROS) access to the hydrophobic tails, and it is reasonable that oxidative attack will preferentially occur at these lower density sites within the membrane. We surmise then that the locations of the observed voids may represent a snapshot of the density fluctuations present in the membrane at the time of ROS exposure. Alternatively, it seems likely that the voids result from the “budding” of clusters of preoxidized lipids which phase-separate from parent lipids within the bilayer. At present, we can not differentiate between the two mechanisms above. Second, the formation of microscopic voids within the supported bilayers represents the appearance of topological defects. Such defects within fluid bilayers are rare because of considerable edge-energy penalty. Two primary factors contributing to the energetics of defect formation within the bilayer environment include reduction in membrane tension energy (-πR2σ, where σ is the membrane tension in dyn/cm2) and an increase in the line-tension penalties (2πRγ, where γ is the line tension in dyn/cm) for a void of dimension R. Because membrane oxidation products may present intrinsic curvatures, they allow for the formation of hemimicellar structure at the
6384
J. Phys. Chem. B, Vol. 114, No. 19, 2010
defect edges and reduce edge energies.41,42 We suspect the observed voids in UV-oxidized lipid bilayers may be stabilized via their accumulation in the micellar edge. However, direct confirmation of this proposition is presently difficult, but emerging chemical characterization techniques, such as spatially resolved, nanoscale secondary-ion mass spectrometry, offer promise.43 It is also instructive to note that the voids we observe appear to be discrete perforations within the otherwise contiguous bilayer. This notion is supported by our observations that by elevating the sample temperature the defects heal. The latter also confirms that the outer periphery of the supported bilayer is pinned, allowing the voids to heal due to the thermal expansion and any thermally induced membrane spreading (or thinning). From the membrane energetic perspective, the stabilization of multiple microscopic voids within the otherwise contiguous perforated bilayers, rather than large macroscopic voids, is rather surprising. Because the total membrane area coverage is practically fixed by the number of initial lipids in the bilayer prior to oxidation on the surface (and macroscopically pinned edges), line tension can be expected to drive the creation of a single (or a few) large void. The stable existence of many voids then must represent a kinetically trapped state of the bilayer. The appearance of multiple voids distributed throughout the surface must then be a result of the formation process, further supporting the notion that the void locations represent a snapshot of the membrane density fluctuations during initial oxidative assaults. Our observations of the thermal cycling of the samples lend additional support to the importance of line tension and the kinetic modulation of void size (see below). The appearance of larger voids after slower cooling (Figure 5D,E) is consistent with the sample approaching thermodynamic equilibrium and the expected limiting case of an infinitely slow cooling rate resulting in the appearance of a large macroscopic void. When samples are heated such that the voids heal over, the fate of the oxidation products is unknown, but some insight can be gained by considering the locations of the voids upon reappearance after cooling. After the sample is held at elevated temperature for a short time, the voids generally appear at or near the original locations. This may be explained by considering the transport of the oxidation products within the membrane. As detailed above, these species may stabilize edges and are likely to preferentially reside in these areas. When the samples are heated such that the voids heal, the coalescence of the edges of the void may result in high concentration of partially oxidized lipids at the location of the former void. Recall that the species have spontaneous curvatures, and hence high concentrations may produce tense and probably nonplanar membrane conformation. When the sample is held at elevated temperature for a short time, either the partially oxidized lipids do not have sufficient time to diffuse away or these “defect” topologies do not have time to relax with the rest of the planar bilayer. On the basis of these considerations, it is reasonable that these regions serve as preferred locations for the reformation of voids upon cooling Third, several factors must be taken into account to explain the long term temporal evolution of voids after the termination of UV exposure. The increased solubility of the membrane oxidation products may contribute to the continued growth of the voids after the termination of UV exposure but is not the only possible mechanism. We first examine the possibility of a continuation of oxidation after illumination. The lifetime of the primary ROS produced in water is short (∼1 min44), and radical propagation is not expected to proceed past the initial monounsaturated lipid. Given the time scales involved, the growth of
Howland and Parikh the voids must not be a result of continuing oxidation and therefore must be due to continued membrane restructuring. We surmise that this restructuring is due to slow desorption of oxidation products generated in situ during the UV exposure. In fact, many independent classes of experiments have revealed the inherent structural incompatibility between single- and twotailed native phospholipids.45,46 In particular, Zhelev and coworkers have demonstrated that forced intercalation of lysolipids into giant unilamellar vesicles results in extended lysolipid desorption. Moreover, using a mass transfer analysis considering adsorption and desorption at the membrane-solution interface as well as transport through a stagnant film, they find that the loss of lysolipids occurs by monomer desorption and follows an exponential decay. Adapting their model, we fit the change in area coverage (as determined by analysis with ImageJ, see Materials and Methods Section for details) of the lipid over time to the following equation
A ) f1 + f2 exp(-kefft)
(1)
where A represents the percent area coverage of membrane on the surface; f1 and f2 are coefficients modified from Zhelev’s model to account for this change; and keff is the effective rate constant for desorption from the membrane. We first find that the observed loss of lipid coverage is well accounted for by this model (Figure 6D). Second, we calculate an average effective rate constant of 8.1 ( 6.7 × 10-4 s-1 from these fits. This value is approximately 2 orders of magnitude lower than the rate observed by Needham and Zhelev47 for monomer desorption from the outer leaflet of vesicles. Direct comparisons between the two experiments are difficult because the effective rate constant includes the mass transfer coefficient for transport through the stagnant layer at the membrane surface. This mass transfer coefficient depends on the geometry of the system and the amount of mixing in the solution phase. In the special case when the transport is limited only by the desorption step, the term bearing the mass transfer coefficient vanishes, and the keff is equal to the true desorption rate. In such a case, we would expect similar rates to be measured in either geometry, if the diffusivities of the two species are equal. The lack of agreement between our values and those measured by Zhelev and coworkers may be due to transport through the stagnant layer or differences in the diffusivity between synthetic lysolipids and the in situ generated species in the current study. Thermal Expansivity and Secondary Material Deposition. Additionally, careful study of the membrane restructuring after ROS attack provides some insight into the bilayer structure and provides the basis for potential applications. First, we can use the change in lipid area coverage as a response to changes in temperature to allow us to calculate the thermal expansion coefficient for the bilayer. The void regions shrink in size in response to a rise in temperature. This shrinkage implies that the macroscopic edges of the bilayer must not extend beyond the edge of the substrate, assuming a positive thermal expansion coefficient. This implies that the bilayer edges must be pinned at the substrate edges, possibly due to substrate adhesion energy. Taking this pinning into account and measuring the observed area coverage change at three different temperatures, we calculate a value of 6.1 ( 2.8 × 10-3/°C for thermal expansion in supported POPC membranes. This value is nominally larger than previously reported values of ∼3 × 10-3/°C in giant vesicles,48 but the two are in general agreement. This suggests that the presence of a substrate does not drastically alter the thermal expansivity of a lipid bilayer.
Membrane Disruption by Oxidative Assault Second, lipid membranes are known to resist the adsorption of many different proteins. The ability to control the size and shape of voids within a membrane offers an attractive platform for directing secondary material deposition. The ability to incorporate secondary species within the patterns of voids within the membrane should generalize easily to directing the deposition of proteins or other soluble molecules of interest. Furthermore, these patterns can be tuned on the fly, possibly allowing for novel configurations of material deposition as the temperature is controlled to precisely tune void size and shape. Acknowledgment. Research supported by the U.S. Department of Energy, Office of Basic Energy Sciences, Division of Materials Sciences and Engineering under Award # DEFG0204ER46173 (A.N.P.) and by a NIH Graduate Fellowship (M. C. H.) via Training Grant in Biomolecular Technology from NIGMS-NIH under Award # T32-GM08799. Supporting Information Available: Figures and captions summarizing control experiments in support of primary experiments described in the main body of the paper above. This material is available free of charge via the Internet at http:// pubs.acs.org. References and Notes (1) Martinet, W.; Kockx, M. M. Curr. Opin. Lipidol. 2001, 12, 535. (2) Spiteller, G. Physiol. Plant. 2003, 119, 5. (3) Hsiai, T.; Berliner, J. A. Curr. Drug Targets 2007, 8, 1222. (4) Cutler, R. G. Proc. Natl. Acad. Sci. U.S.A. 1985, 82, 4798. (5) Markesbery, W. R. Free Radical Biol. Med. 1997, 23, 134. (6) Borchman, D.; Lamba, O. P.; Salmassi, S.; Lou, M.; Yappert, M. C. Lipids 1992, 27, 261. (7) Megli, F. A.; Russo, L. Biochim. Biophys. Acta, Biomembr. 2008, 1778, 143. (8) Kaplan, P.; Racay, P.; Lehotsky, J.; Mezesova, V. Neurochem. Res. 1995, 20, 815. (9) Ayuyan, A. G.; Cohen, F. S. Biophys. J. 2006, 91, 2172. (10) Samsonov, A. V.; Mihalyov, I.; Cohen, F. S. Biophys. J. 2001, 81, 1486. (11) Jacob, R. F.; Mason, R. P. J. Biol. Chem. 2005, 280, 39380. (12) Megli, F. M.; Russo, L.; Sabatini, K. Febs Lett. 2005, 579, 4577. (13) Vercesi, A. E.; Kowaltowski, A. J.; Grijalba, M. T.; Meinicke, A. R.; Castilho, R. F. Biosci. Rep. 1997, 17, 43. (14) Greenberg, M. E.; Li, X. M.; Gugiu, B. G.; Gu, X. D.; Qin, J.; Salomon, R. G.; Hazen, S. L. J. Biol. Chem. 2008, 283, 2385. (15) Watson, A. D.; Leitinger, N.; Navab, M.; Faull, K. F.; Horkko, S.; Witztum, J. L.; Palinski, W.; Schwenke, D.; Salomon, R. G.; Sha, W.; Subbanagounder, G.; Fogelman, A. M.; Berliner, J. A. J. Biol. Chem. 1997, 272, 13597.
J. Phys. Chem. B, Vol. 114, No. 19, 2010 6385 (16) Johnson, J. M.; Ha, T.; Chu, S.; Boxer, S. G. Biophys. J. 2002, 83, 3371. (17) Keller, C. A.; Glasma¨star, K.; Zhdanov, V. P.; Kasemo, B. Phys. ReV. Lett. 2000, 84, 5443. (18) Bayerl, T. M.; Bloom, M. Biophys. J. 1990, 58, 357. (19) Johnson, S. J.; Bayerl, T. M.; McDermott, D. C.; Adam, G. W.; Rennie, A. R.; Thomas, R. K.; Sackmann, E. Biophys. J. 1991, 59, 289. (20) Castellana, E. T.; Cremer, P. S. Surf. Sci. Rep. 2006, 61, 429. (21) Sanii, B.; Parikh, A. N. Annu. ReV. Phys. Chem. 2008, 59, 411. (22) Szoka, F.; Olson, F.; Heath, T.; Vail, W.; Mayhew, E.; Papahadjopoulos, D. Biochim. Biophys. Acta 1980, 601, 559. (23) Cremer, P. S.; Boxer, S. G. J. Phys. Chem. B 1999, 103, 2554. (24) Tamm, L. K.; McConnell, H. M. Biophys. J. 1985, 47, 105. (25) Howland, M. C.; Szmodis, A. W.; Sanii, B.; Parikh, A. N. Biophys. J. 2007, 92, 1306. (26) Axelrod, D.; Koppel, D. E.; Schlessinger, J.; Elson, E.; Webb, W. W. Biophys. J. 1976, 16, 1055. (27) Kucerka, N.; Tristram-Nagle, S.; Nagle, J. F. J. Membr. Biol. 2006, 208, 193. (28) Recall that the parent bilayer has been exposed to UV resulting in a drastic reduction in fluoresence intensity. Bright features are consistent with the addition of new, unbleached molecules. (29) Ratto, T. V.; Longo, M. L. Biophys. J. 2002, 83, 3380. (30) Sanii, B. Frappe Software. (31) Yee, C. K.; Amweg, M. L.; Parikh, A. N. AdV. Mater. 2004, 16, 1184. (32) Smith, H. L.; Howland, M. C.; Szmodis, A. W.; Li, Q. J.; Daemen, L. L.; Parikh, A. N.; Majewski, J. J. Am. Chem. Soc. 2009, 131, 3631. (33) Kinder, R.; Ziegler, C.; Wessels, J. M. Int. J. Radiat. Biol. 1997, 71, 561. (34) Santrock, J.; Gorski, R. A.; Ogara, J. F. Chem. Res. Toxicol. 1992, 5, 134. (35) Wentworth, P.; Jones, L. H.; Wentworth, A. D.; Zhu, X. Y.; Larsen, N. A.; Wilson, I. A.; Xu, X.; Goddard, W. A.; Janda, K. D.; Eschenmoser, A.; Lerner, R. A. Science 2001, 293, 1806. (36) Bhamidipati, S. P.; Hamilton, J. A. Biochemistry 1995, 34, 5666. (37) Fuller, N.; Rand, R. P. Biophys. J. 2001, 81, 243. (38) Bayerl, T. M. Curr. Opin. Colloid Interface Sci. 2000, 5, 232. (39) Dixit, S. S.; Szmodis, A.; Parikh, A. N. Chemphyschem 2006, 7, 1678. (40) Boulbitch, A. Europhys. Lett. 2002, 59, 910. (41) Karatekin, E.; Sandre, O.; Guitouni, H.; Borghi, N.; Puech, P. H.; Brochard-Wyart, F. Biophys. J. 2003, 84, 1734. (42) Mills, J. K.; Needham, D. Biochim. Biophys. Acta, Biomembr. 2005, 1716, 77. (43) Kraft, M. L.; Weber, P. K.; Longo, M. L.; Hutcheon, I. D.; Boxer, S. G. Science 2006, 313, 1948. (44) Lerner, R. A.; Eschenmoser, A. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 3013. (45) Zhelev, D. V. Biophys. J. 1996, 71, 257. (46) Needham, D.; Stoicheva, N.; Zhelev, D. V. Biophys. J. 1997, 73, 2615. (47) Needham, D.; Zhelev, D. V. Ann. Biomed. Eng. 1995, 23, 287. (48) Evans, E.; Needham, D. J. Phys. Chem. 1987, 91, 4219.
JP102861V
